Doctoral Research Project: " Modelisation of the oceanic thermohaline circulation using Planetary Geostrophic equations. "
The time and space scales used for climatic studies involving the ocean thermohaline circulation allows an interesting simplification in the traditionnal primitive equations set. For time scales much longer than a day and horizontal scales larger than the Rossby radius of deformation, one can get rid of time derivatives and non-linear terms in the momentum equations. Then, if the density field and the surface windr-stress are known, the velocities are just diagnostic. The full prognostic equations are conserved for the evolution of tracers fields, temperature and salinity. This filters out all fast propagating waves, and the only baroclinic Rossby waves remains : consequently, the time step used for simulations can be much longer than the ones necessary in PE models, mainly when the atmospheric forcing is time dependent.
The choice of the momentum dissipation parameterization is then a necessary step in order to include the sub-grid-scale transfer of energy and to match boundary conditions. Although Laplacian operators are commonly used there, because similar to molecular-scale viscosity and 'physically' consistent, various alternative are implemented: linear Rayleigh friction is much easier and quicker to implement, but also biharmonic dissipation or no dissipation at all. Different associated boundary conditions are tested: the common no-slip, but also free-slip and no-normal-flow boundary conditions using a vorticity closure to compute the tangential velocities along the boundaries. A fully 3-d linear friction relaxing the hydrostatic approximation is also used. The question we consider in the following is: in what extent can boundary layers structure influence the large-scale ocean circulation ?
The influence of these momentum dissipation parameterizations and associated boundary conditions on the ocean thermohaline circulation is analyzed in a coarse- resolution (160km) mid-latitude box-geometry beta-plane forced by restoring boundary conditions for the surface temperature. No wind forcing is used here, since our main concern is the baroclinic thermohaline circulation and not the thermohaline effect of the barotropic wind-driven gyres. In short, our main conclusions are that the whole large-scale ocean circulation is strongly influenced by momentum dissipation and boundary conditions. The different models we have run generated bottom water temperatures varying from .5 °C, meridionnal overturning streamfunction almost doubling from 8 Sv to 15, but quite uniform feedbacks to the atmosphere in terms of polar heat transport or zonally integrated surface heat fluxes. The details of the dynamic that are responsible for such large scale diagnostics variability involve mainly the lateral boundaries and the boundary conditions on lateral walls. We show that Laplacian friction associated with no-slip boundary conditions generates strong vertical velocities that modify the whole basin mass balance. On the other hand, a vorticity closure that limits the vertical velocities by allowing horizontal recirculation along the boundaries generate much less noisy vertical velocity fields, which become more correlated to the deep convection patterns. Upwelling at the base of the (thermally driven) Gulf Stream, the so-called 'Veronis effect', is efficiently reduced and the whole overturning circulation, even if much weaker than with no-slip boundary conditions, produce cooler deep water and highr polar heat transport.
Most of these models are very fast to integrate (1h CPU on an IBM/RISC6000 for a three thousand year simulation on a 32x28x15 grid) and allow extensive studies of parameter spaces. We also used them for studying spontaneous decadal oscillations under fixed heat flux boundary conditions, either zonally uniform or diagnosed from an initial run under restoring surface boundary conditions. The sensitivity of these oscillations to the various model processes was thoroughly analyzed to help unravel the driving mechanism: Longwave baroclinic instability of the mean circulation, in the region of the western boundary current separation from the coast and its eastward extension, was found to sustain the oscillations against dissipation. This was the highlight of my Ph.D. thesis research that pushed me in the field of climate variability.
Postdoctoral Research Project
Because of the rather unusual mechanism found for such large scale oscillations, the robustness of this variability was assessed against several factors: model resolution, especially in eddy-resolving cases, coupling to simplified energy balance atmospheric models, wind forcing and realistic bottom topography. This work confirmed the plausibility of such an ocean mode to explain the observed Atlantic multidecadal Oscillation.
CNRS (French National Center for Scientific Research) Research Project